Singapore's relative humidity swings by about thirty points between dawn and mid-afternoon. Almost none of that is water arriving or leaving.
It is the thermometer moving. Relative humidity is a ratio whose denominator depends on temperature, so quoting it without the temperature beside it describes very little.
What the ratio is a ratio of
The World Meteorological Organization defines relative humidity as the observed vapour pressure divided by the saturation vapour pressure at the same temperature and pressure, expressed as a percentage. The denominator climbs steeply with warmth — roughly a doubling for every eleven to twelve degrees in the tropical range, and slightly faster in cooler air.
This leads to a common, and incorrect, comparison:
| London, 15 °C, 70% | Singapore, 31 °C, 70% | |
|---|---|---|
| Saturation vapour pressure | 17.0 hPa | 44.9 hPa |
| Actual vapour pressure | 11.9 hPa | 31.4 hPa |
| Water per kilogram of dry air | 7.4 g | 19.9 g |
| Dew point | 9.6 °C | 24.9 °C |
The same percentage, two and a half to two and seven tenths times the water. The range exists because you can measure the difference per kilogram of dry air or per cubic metre, and warm air is less dense; either way the figure is nowhere near equal.
But the popular correction can overshoot. Relative humidity is not useless, just incomplete; paired with its temperature, it contains the same information as the dew point, one substitution away. The heat index proves the point: the formula American forecasters use takes relative humidity as an input and works, precisely because it takes the temperature too.
A Singapore day, computed from the met service's own normals
The national met service publishes the diurnal pattern plainly: relative humidity varies from more than 90% just before sunrise to around 60% in the afternoon on days without rain, with a mean around 82%. It also publishes the temperature normals for the same hours. Putting the two together:
| Month | Pre-dawn | Mid-afternoon | Change in RH | Change in water content |
|---|---|---|---|---|
| January | 24.3 °C, 94.7% | 30.6 °C, 66.0% | −28.7 points | +0.8% |
| April | 25.3 °C, 95.6% | 32.4 °C, 62.3% | −33.3 points | −1.7% |
| August | 25.3 °C, 93.0% | 31.4 °C, 63.6% | −29.4 points | −2.6% |
| October | 25.0 °C, 95.2% | 31.8 °C, 61.4% | −33.8 points | −4.3% |
Thirty-odd points of relative humidity, under five per cent of actual water. The dew point barely moves — under a degree in every month above.
Kuala Lumpur shows the same thing hour by hour. Take the 6am air at 26.6 °C and 80%, warm it to the afternoon's 31.4 °C and change nothing else, and the arithmetic gives 60.6% relative humidity. The observed afternoon figure was 60%. The entire twenty-point swing is the thermometer.
This explanation is derived from the data, not published by the meteorological services themselves. They publish the temperatures and the humidities; the dew points and water contents above are arithmetic on their numbers, shown so they can be redone. The mechanism itself is stated by the American weather service, which notes that with dew point unchanged, relative humidity is highest in the early morning when air is coolest and lowest in the afternoon when it is warmest.
The number that travels
Dew point is the temperature to which air must be cooled, at constant pressure, to reach saturation. Because it tracks the water itself rather than a ratio, it can be compared between places and hours without carrying a second number alongside.
The American weather service publishes two thresholds for summer comfort: at or below 55 °F, about 12.8 °C, it calls the air dry and comfortable; at or above 65 °F, about 18.3 °C, there is enough moisture that conditions become oppressive.
Those two are the only bands I could source to a meteorological authority. The five- and seven-row "dew point comfort scales" that circulate widely appear only on commercial and hobbyist sites, and given that Singapore's dew point sits close to 24 °C for much of the year — well past the upper threshold — a finer scale would not tell a reader here much anyway.
Why a dehumidifier's rating is not a promise
This confusion has financial consequences, as an American appliance-testing programme demonstrates. When the test procedure for portable dehumidifiers changed, the test room went from 80 °F to 65 °F — and the programme states the consequence directly, that a unit tested to the new procedure will report a smaller capacity than under the old one.
The programme's conversion table shows that a machine formerly rated for 30 pints a day is now rated for about 20. One formerly rated 60 is now about 35. Nothing about the machine changed. The room got fifteen degrees cooler, and the number on the box fell by roughly two fifths.
Capacity ratings are national, and they do not agree. The American procedure now tests at 65 °F and 60% relative humidity; the Chinese standard quotes a nominal capacity at 27 °C and 60%, and a rated capacity at 30 °C and 80%. There is no international performance standard — the international electrical standard that covers dehumidifiers addresses safety, not capacity.
This creates a regulatory gap in the ASEAN market worth knowing about. Singapore's mandatory energy labelling and performance standards cover air conditioners, refrigerators, clothes dryers, televisions, lighting, water heaters and several commercial categories. Dehumidifiers are on neither list. So a unit sold here carries whichever foreign rating its vendor chose to print, and the warmest, wettest test conditions produce the largest number.
Three Common Beliefs Disproved by the Numbers
That 100% means rain is false, and the weather service says so explicitly: at saturation, precipitation does not necessarily occur. The local version is sharper still, since the Singapore met service records that humidity frequently reaches 100% during prolonged rain — saturation at ground level follows the rain rather than causing it.
That humidity makes heat worse because the air traps heat is the wrong mechanism. What humid air does is narrow the vapour-pressure difference across your skin, which is what drives sweat to evaporate. The heat index is built on exactly that: sweating rate, skin resistance to moisture transfer as a function of vapour-pressure difference, and ventilation rate. It is an evaporation model, not a heat-retention one.
And "the rain will break the humidity" is the tropical cousin of "it's too cold to snow"—a plausible-sounding claim that gets the physics backwards. An afternoon storm dropping the temperature from 32 °C to 25 °C with the dew point unchanged at 23 °C takes relative humidity from about 62% to about 88%. The air is identical; it is merely cooler, and the number went up.
Reading a forecast after this
When a figure is quoted without a temperature, it is describing a ratio whose denominator you have not been told. Ask for the dew point, or ask for the temperature that goes with the percentage — either completes it.
And when buying anything that removes water from air, the question is what conditions its capacity was measured at. On the evidence above, that single question is worth about forty per cent of the number on the box.
Sources, and what is arithmetic rather than testimony
The definitions of relative humidity, mixing ratio and dew point, and the saturation vapour pressure formula with its stated validity range, are from the World Meteorological Organization's guide to instruments and methods of observation, number 8, annexes 4.A and 4.B. The Singapore diurnal pattern, the mean annual figure and the statement about rain are from the Meteorological Service Singapore's climate page, with the monthly temperature and humidity normals for 1991–2020 from the same source; the Kuala Lumpur observations are from the Malaysian Meteorological Department's station pages. The comfort thresholds, the mechanism behind the diurnal swing and the correction on saturation are from the United States National Weather Service; the heat index formula and its basis from the Rothfusz technical attachment of 1990. The dehumidifier test conditions and the conversion table are from the ENERGY STAR programme's own page on testing and capacity, and the Singapore labelling scope from the National Environment Agency. All were read on 20 September 2026.
The dew points, mixing ratios and percentage changes in this piece are my arithmetic on published temperature and humidity, not figures either met service publishes — neither publishes dew point at all, and neither states the mechanism. The formula and the working are given above so the numbers can be checked rather than taken. Two figures are held loosely: the Chinese rating conditions come from technical summaries rather than the standard itself, which could not be retrieved, and the often-quoted seven per cent per degree for atmospheric moisture is a global-mean figure anchored near freezing — in tropical air the local rate is closer to six, which is why the doubling interval above is given as eleven to twelve degrees rather than ten.
